Animal Spatial Cognition

The following sections are included:

• PREFACE

• ACKNOWLEDGEMENTS

• CONTENTS

The survival of many species depends on their ability to use spatial knowledge. Rodents searching for food far from their burrow must escape the shadow of the predator or they will be caught. A pigeon released several hundred miles from its loft must be able to find the right direction home, otherwise it will get lost and die. In this hard and hostile world, animals must know where to find places where food is plentiful and keep in mind their homeward direction in case of approaching danger. This helps us understand why spatial abilities are well developed in many animal species, and have, for some animals such as homing pigeons and bees, become highly sophisticated. Thus it is quite understandable why animals’ spatial representations have been debated through the five decades from the infancy of experimental psychology, and why they are still being investigated in many studies on a wide range of species.

All mammals react in some way or other to novelty. Exploratory behaviors are particularly obvious in rodents. A rat or a hamster placed in a new environment displays feverish investigatory activity, which finds expression in multiple contacts with objects in the environment, sniffing, stopping, rearing up on the hind paws, and so on. In human subjects, reactions to novelty, although different, are also observable. Young children displaying an irrepressible need to seize and touch any unfamiliar object within their reach, people entering a restaurant and looking around with discretion, provide convincing illustrations. One feature of this activity is its decrease over time, a phenomenon called habituation. Depending on the complexity of the situation, its degree of novelty, and the species concerned, the process of habituation varies.

Since Tolman cast the cognitive map hypothesis, many studies have been conducted, especially since the nineteen-seventies, in order to provide it with experimental support. As in Tolman’s experiments, most of the situations aimed at investigating the nature of spatial representations in rodents are multi-trial learning tasks requiring the subjects to locate and reach a generally invisible goal in a maze or an arena. Indeed, the animals must be motivated enough to complete the task. All the paradigms involve, therefore, a reinforcing event encouraging animals to implement a proficient strategy for reaching the goal either because they will find a reward there, or, alternatively, a safe place to escape from an uncomfortable medium (such as from cold water in the Morris swimming pool). Thus, these situations differ markedly from those described in the preceding chapter, which generally take place within a relatively short period of time, rely on spontaneous behavior, and (with the exception of the problem-solving tests) do not involve any explicit reinforcement. The fact that animals are repeatedly exposed to the same situation also represents a consistent difference with the formerly described paradigms. Animals subjected to a training experience may easily identify, trial after trial, the spatial relationships existing among some locations - the starting point, the choice points or the goal - which provide different views of the experimental layout, and have a particular status in the picking up of information.

The mere ability to negotiate a detour, i.e., to deviate from a direct path toward a goal as soon as an obstacle is perceived, was considered by Kohler (1927) as indicative of “intelligent behaviors”. Some species, such as hens are unable to head directly toward and around the end of an obstacle that separates them from food, persistently attempting to pass “through” the obstacle. Other studies that have been carried out to demonstrate similar effects have examined the animals’ ability to take advantage of favorable alternatives; for example, the availability of a new pathway which is shorter than the usual one, allowing the animal to get to a goal more quickly by taking a shortcut.

In parallel with the empirical study of orientation and map construction, using studies such as those described in the foregoing chapters, theoretical models have also been developed. Models are theoretical constructs which attempt to incorporate and fit together as many pieces (experimental data) of the jigsaw puzzle (spatial cognition) as possible. In my opinion, they should not be regarded as describing the process itself but rather what it could be. Even if they do not always explicitly lead to experimentally testable predictions, they have an invaluable heuristic function.

A major event in the field of spatial cognition and its neurobiological bases has been the discovery of a population of hippocampal neurons, the properties of which reflect unambiguously the encoding of the animal’s location in the environment. They were christened “place-cells” by O’Keefe and Dostrovski who discovered them in 1971. Despite some debates about the precise meaning of place-cell firing properties, and also about the involvement of such cells in non spatial processing, this finding has given a strong impetus to research in the domain of the brain substrates of spatial cognition. A few years later, O’Keefe and Nadel (1978) proposed that the hippocampus represented the neural substrate of cognitive maps. As already discussed in Chapter 1, the integration of findings about brain functioning and spatial behavior into a unified theory strengthened the impact of the initial discovery, giving rise to a growing interest in rodents’ hippocampus and spatial memory. Some years earlier, the observation by Scoville and Milner (1957) of the dramatic temporally graded amnesia produced in H.M. by bilateral lesions to the hippocampus and related temporal lobe structures had a comparable general impact in the field of human memory. This chapter is aimed at giving an overview of studies which allow us to better understand the role played by the hippocampus and associated structures. Studies of the effects of lesions on traditional learning tasks, exploratory behavior and spatial problem solving tests, are presented in the first two sections. The other sections are devoted to electrophysiological studies.

It is unquestionable that the hippocampus plays a key role in spatial processing. Examples of actual spatial performance that are spared by hippocampal lesions are rare. The deficits are usually pronounced and unrecoverable. Moreover, data from electrophysiological studies add further support for a spatial function for the hippocampus. Does that mean that the hippocampal formation is the unique site in the brain where spatial processing is achieved? The question can be formulated in another way: are there aspects of spatial processing that are not ensured by the hippocampus but that are necessary in order that the whole process can take place, from the very first beginnings of exploration up to the point where orientation is based on coherent, effective representations? It appears difficult to maintain that a single structure, what is more, a paleocortical structure, acts alone at all the various stages of the complex spatial process. It is common sense to assume otherwise.

Besides the associative parietal area, other neocortical regions, the prefrontal cortex and, to a lesser extent, the occipital cortex, appear to be involved in spatial processing. In spatial tasks, damage to the medial prefrontal cortex has been shown to result in an apparently heterogeneous pattern of effects, which vary according to task requirements. However, a good deal of data supports the idea that this region is involved in spatial behavior. The picture is far less clear concerning the occipital region of the neocortex classically considered to be exclusively responsible for the processing of visual features. However, some lesion data suggest that this region could also be involved in spatial processing, independently of the specific impairment of visual form perception. This idea is not new since Lashley (1943) and Tsang (1934, 1936) had found severe deficits in rats with destruction of the occipital area when they were tested on various spatial tasks. De Renzi (1982) in his book “Disorders of Space Exploration and Cognition” quotes several early clinical observations of patients with occipital damage who displayed orientational deficits (Foerster, 1890; Meyer, 1900; Peters, 1896), though these were casual observations the conclusions of which have not been investigated further.

One of the aims of this chapter is to present models which deal with the implementation by the brain of spatial processing and memory. Some of these models exclusively concern the hippocampus, either because they account only for one specific type of encoding (setting up of place-cell firing fields: Sharp; distance encoding: Muller and coworkers), or because the basic assumption is that the hippocampus can perform all of the stages in the process (O’Keefe). Other conceptions rely on the cooperation of the hippocampus with neocortical structures, such as the parietal cortex (Poucet; McNaughton and coworkers). Finally, four general models of memory are presented (Teyler and DiScenna; Miller; Damasio; McClelland and coworkers). Although these models are not specifically aimed at accounting for spatial processing, these are of interest in the present context: in Chapters 7 and 8, I have presented several data which strongly suggest that the associative parietal and prefrontal cortices, and to a lesser extent the occipital area, do subserve complementary spatial functions. Consequently, it is of importance to have some insight, or, at least, hypotheses about how their possible interactions and cooperation with the hippocampal system may be implemented within the brain. In this respect, theories concerning the neural substrates and evolution over time of memories in general provide interesting perspectives which can be applied to spatial memories in particular. Indeed, the memory components of spatial processing can be reasonably thought to be controlled by the same brain mechanisms as those involved in the memory components of any other type of processing (of comparable cognitive level).

I drew attention in the first chapter to the fact that the term “cognitive map” could be misleading because it conveys the notion of a static entity. Throughout the present book, the majority of questions that have been raised concern the interpretation of data in relation to the construction and use of cognitive spatial maps: to what extent do such-and-such results provide evidence that animals possess cognitive maps? Are there more parsimonious hypotheses to account for the observed behaviors? And so on. Some of these questions arise from the fact that the “cognitive map” notion is seldom taken as a metaphor, but more often stricto sensu. Loosely defined by Tolman, often unwittingly used by many authors, although more precisely circumscribed by O’Keefe and Nadel (1978), the term is not satisfactory. A first imperative should be to propose an alternative term for high level spatial representations in animals. I would propose the term “spatial integrating system” as a suitable replacement, since this term emphasizes an aspect which had been stressed by Tolman but more or less neglected by present day researchers, namely the dynamic aspect of spatial representations. In my view, taking into account this property of so-called cognitive maps may assist the better understanding in the future of their underlying nature and properties, and the setting up of new experimental hypotheses.

The following sections are included:

• REFERENCES

• SUBJECT INDEX